[3K1-IS-3] Agents

[3K1-IS-3-05] International privacy and data protection law to the test of agentic AI

〇William Axel Letrone¹ (1. CNRS, Nantes University (France))

キーワード：Agentic AI, International law, privacy

Agentic AI is about to experience large diffusion in society. Agentic AI may be understood as an AI system capable of executing a diverse suite of more-or-less complex tasks without human involvement. Agentic AI systems are designed to exhibit high levels of autonomy, functioning with minimal human inputs. However, agentic AI raises privacy and data protection issues that need to be addressed before the technology becomes ubiquitous.
(1) Objective
This prospective research anticipates the generalization of autonomous AI agents within society. As such, it aims to analyze agentic AI from the perspective of international privacy and data protection law, drawing upon widely-recognized data protection principles, in order to assess whether current frameworks on privacy are up to the task of regulating agentic AI systems.
(2) Results
The research will provide well-needed clarifications on the categorization of agentic AI systems under the AI Act and the consequences of their categorization. It will also provide relevant legal insights to enlighten the application of key privacy and data protection principles, such as data minimization and purpose limitation.
(3) Method
The research will employ an evaluative legal research methodology, where key legal rules are mobilized to assess their applicability in new, concrete contexts.